Lentis/Online Consumer Reviews

An online consumer review is a review posted online about a product or service, usually written by a customer who has previously purchased the product or service. The customer posts this review on the business’ website to provide feedback about the product and inform other customers before they purchase the same product. These reviews influence the decisions of other customers with about 82% of Americans saying they sometimes look at online reviews before making a purchase and about 40% saying they almost always look at reviews before making a purchase [1].

These reviews are not always trustworthy though, as fraudulent reviews are a growing problem. Some online businesses will pay for positive reviews on their own products or poor reviews on competing products in order to influence customers’ decisions [2].


Introduction

History

In the mid-1990s, major e-commerce websites such as Amazon.com, Deja.com, and Cnet began including online customer product reviews [3]. Epinions.com, founded in 1999, was the first major consumer review website [4]. It changed the landscape for online reviews in several important ways. It was the first third-party consumer review website which freed it from the accusations of bias faced by product reviews on e-commerce websites that sell those products [4].  Furthermore, it paid consumers to review a wide variety of products. Before this, product reviews were written primarily by professionals (for example, the Consumer Reports magazine which has been testing and reviewing products since 1936 [5]).

Initially, Epinions.com faced skepticism of the usefulness and credibility of consumer reviews [4]. However, it paved the way for popular review sites such as Yelp, founded in 2004, and TripAdvisor, founded in 2000. These sites work on social recognition rather than a cash incentive. Yelp, which provides consumer reviews of local businesses, now has over 171 million reviews on its site [6].  TripAdvisor, which provides consumer reviews of travel destinations, now has over 661 million reviews [7]. Today, two thirds of online shoppers find customer reviews trustworthy and always or sometimes make decisions based on these reviews [8]. 84% of consumers trust online reviews as much as personal recommendations [1].

Scope of the Problem

Ben Zhao at the University of Chicago has researched the phenomena of fake online consumer reviews. He has indicated two methods used by fake reviewers: crowdturfing campaigns and artificial intelligence.

Crowdturfing is combination of “crowdsourcing” and “astroturfing.” Crowdturfing occurs when "bad actors pay groups of users to perform questionable or illegal actions online" [9]. In terms of online reviews, humans are recruited to write fake consumer reviews to support or oppose certain products. These reviews often go undetected by automated programs because they are written by real humans. The scope of these campaigns is limited by the necessary compensation of the writers. This gives larger companies an advantage over smaller companies.

Artificial intelligence (AI) has also been used to create fake online consumer reviews. Basic computational models have been programmed in the past to post fake online consumer reviews, however, the span of their ability to resemble reviews written by real humans is lacking. As this technology has improved, neural networks have become more popular. Neural networks are computational models that use a connected network of neurons to solve machine learning tasks [9]. These 'neurons' are databases full of information about online consumer reviews: common words/phrases, sentence structure, etc. The neural networks learn how to write reviews that resemble those written by a real human thus going undetected by automated programs as well.


Case Studies

Amazon

After concluding that customers distrust paid reviews, Amazon banned paid reviews and now claims to police its site for these reviews. However, banning fraudulent reviews hasn’t eliminated them but has forced companies to work more secretly so these fake reviews are harder to identify. Fraudulent reviews have caused problems not only for Amazon, but also for companies that do not employ fraudulent reviews. Tommy Noonan, operator of ReviewMeta, a website that helps consumers identify fake Amazon reviews, said, “These days it is very hard to sell anything on Amazon if you play fairly. If you want your product to be competitive, you have to somehow manufacture reviews,” [10].

Facebook has become a medium for sellers to form groups and seek out shoppers to write fake reviews for them in return for money or some other form of compensation. These groups use techniques to avoid being detected by Amazon, such as targeting shoppers with histories of writing Amazon reviews. Shoppers will often agree to write fake reviews simply as a way to make a little money [10]. In one instance, a shopper purchased an item from Amazon because it had great reviews, only to receive the item and discover that the cheap plastic didn’t serve its purpose at all. Despite having fallen victim to the result of fake reviews, this same customer now writes fake reviews for companies on Amazon. When asked why, he replied, “I don’t think it’s right that people can write fake reviews...but I need the money,” [11].

Many of the companies using fraudulent reviews on Amazon are from China, looking to enter the US market. In using Amazon to sell their products, these companies can use Amazon’s reputation to get business from customers. This could eventually be harmful to Amazon as customers begin to realize the poor quality of their purchases. Despite Amazon’s efforts to eliminate fake reviews, there are too many groups out there selling reviews secretively to ever be eliminated completely. As Amazon’s algorithms improve, reviewers change their procedures to reduce suspicion. As long as there is an incentive for fake reviews, they will continue to be a problem [11].

The Shed at Dulwich

In April 2017, Oobah Butler, a reporter for Vice, created a fake restaurant and achieved the ranking of #1 in London on TripAdvisor by falsifying customer reviews. He had previously written fake reviews for restaurants on TripAdvisor, finding his clients via freelance job postings. This inspired him to test the effectiveness of fake reviews by turning his garden shed in Dulwich, London into a fake restaurant. He called it the “The Shed at Dulwich” [12].

Butler purchased a burner phone, photographed plates of fake food, and recruited friends to write five star reviews of his restaurant on TripAdvisor. The restaurant started out ranked #18,149 - the lowest ranked restaurant in London on TripAdvisor [13]. However, as it received more reviews, it continued to climb the rankings until 6 months later, on November 1st, 2017, it became the #1 restaurant in London [13]. One review stated “Spent a weekend in London and heard through the grapevine that this place is a must-visit. After a few mildly frustrating phone calls, I was in.” [14] To increase credibility, one reviewer included negative feedback, stating that they were offered a blanket with a stain, but still rated The Shed at Dulwich five stars [14].

The restaurant was so successful it even got a one star review, likely from a rival restaurant. Ironically, TripAdvisor identified this review as fake and removed it [12]. Butler was eventually fielding hundreds of calls for reservations on his burner phone [13]. By making it difficult to get a reservation, Butler made the "restaurant" exclusive which further increased demand.

TripAdvisor’s software scans reviews for strong positive or negative bias to catch fakes. It also tracks the IP addresses of the reviewers for suspicious activity [15]. However, Butler’s method of employing friends to write realistic reviews from a variety of computers made the falsehood virtually undetectable.

This case demonstrates both the power and the fragility of online consumer reviews. Online consumer reviews have the ability to make or break a business and yet the system is highly vulnerable to false information, even on one of the most trusted review websites. The Shed at Dulwich was eventually identified as fake and was removed from TripAdvisor’s site. Butler, however, remains skeptical that TripAdvisor is closer to a solution for the fake review problem. He said “Put it this way: I’m still not banned from TripAdvisor” [12].


Detecting Fake Reviews

The increase in fake online consumer reviews has instigated the creation of a fake review spotting website: fakespot.com. This website uses artificial intelligence to determine the trustworthiness of reviews on websites such as Amazon, TripAdvisor, Yelp and Walmart [16]. Consumers can use it while searching through reviews, but it may not always be reliable. Consumers should learn how to determine fake reviews on their own.

Be Skeptical

Consumers should trust themselves to be able to spot fake online reviews. Every online review should be examined closely before it is believed.

Check the Dates

Crowdturfing campaigns will have their writers post their fake reviews around the same time to have a bigger impact on the most recent reviews. If there is a large surge in reviews around the same time frame and they all say similar things, they could be fake.

Consider Language

If reviews are difficult to understand or completely incomprehensible, then they are likely to be fake reviews.

Examine Profiles

Accounts are used to create online consumer reviews. To determine if a review is fake, consumers can examine the reviewers profile. If the profile looks fake, the review could be fake as well. Consumers can also google reviewers profile to determine whether or not the reviewers can be trusted.

Verify Purchases

Companies that sell products will often verify when a reviewer has actually purchased the product. If this is the case, the reviews have more credibility.


Conclusions

People trust the opinions of their peers and often make decisions based on these opinions. Companies can take advantage of this system by falsifying reviews. Customers should therefore critically evaluate online reviews before making purchasing decisions. Further research could examine the effect of falsified reviews on the credibility of online vendors and the effectiveness of the methods that websites use to detect fake reviews.


Sources

  1. http://www.pewinternet.org/2016/12/19/online-reviews/
  2. http://www.digitalethics.org/essays/trend-toward-falsifying-online-reviews
  3. https://www.nytimes.com/1999/07/11/magazine/instant-company.html
  4. https://www.forbes.com/sites/ericgoldman/2014/03/12/epinions-the-path-breaking-website-is-dead-some-lessons-it-taught-us/
  5. https://www.chicagotribune.com/news/ct-xpm-2000-05-01-0005010029-story.html
  6. https://www.yelp.com/about
  7. http://ir.tripadvisor.com/static-files/a8b88a72-0d45-47a6-be65-e47de05141b5
  8. https://www.nielsen.com/content/dam/nielsenglobal/apac/docs/reports/2015/nielsen-global-trust-in-advertising-report-september-2015.pdf
  9. https://arxiv.org/pdf/1708.08151.pdf
  10. https://www.washingtonpost.com/business/economy/how-merchants-secretly-use-facebook-to-flood-amazon-with-fake-reviews/2018/04/23/5dad1e30-4392-11e8-8569-26fda6b404c7_story.html
  11. https://www.npr.org/2018/07/30/629800775/some-amazon-reviews-are-too-good-to-be-believed-theyre-paid-for
  12. https://www.newyorker.com/tech/annals-of-technology/the-never-ending-war-on-fake-reviews
  13. https://www.vice.com/en_us/article/434gqw/i-made-my-shed-the-top-rated-restaurant-on-tripadvisor
  14. https://www.washingtonpost.com/news/food/wp/2017/12/08/it-was-londons-top-rated-restaurant-just-one-problem-it-didnt-exist/
  15. https://www.tripadvisor.com/TripAdvisorInsights/w3688
  16. fakespot.com
  17. https://www.realsimple.com/work-life/life-strategies/fake-online-reviews